't

, A nunlerical study, of bench blast row delay timing and its influence on percent-cast* _'z'} _//3 _: _ . .....
._ r_
.....i:_'_::_
_ _''_'_

Dale S, Preece, Sandia National Laboratories, Albuquerque, New Mexico, USA ¢_,_

Abstract: The computer program, DMC (D_istinct Motion Code), which was developed for simulating the rock motion _ "_
associated with blasting, has been used to study the influence of row delay timing on rock motion. The numerical simu- _ _

lations percent-cast
lower correspond with
than field observations
a medium in that
delay (100 very ms).
to 200 shortThe
delays
DMC(< predicted
50ms) andrelationship
very long delays (> 300ms)
between produce
row delay timinga ._-_'_
._
and percent-cast is more complex than expected with a dip in the curve where the optimum timing might be expected. _ _ _
More study is required to gain a full understanding of this phenomenon, a, _

1 iNTRODUCTION is currently being used on a SUN SPARCstation 10-41

computer workstation.manufactured by Sun Microsys-
The largest mining operations in the world are surface tems Incorporated.
coal mines which involve processing and movement of DMC with the coupled gas flow capability has been
significant amounts of rock and soil to extract the coal exercised on a bench blast configuration with different
and reclaim the land. The surface coal mining industry delay times between rows. This exercise was performed
has adopted a blasting technique called Cast Blasting to test the ability of the code to predict differences in
which allows mining engineers to move a maximum blast-induced rock motion over a wide range of row delay
amount of overburden material with explosives. This times. Practical blasting experience indicates that very
leaves less material to be moved with mechanical equip- short delay times (< 50 ms) as well as very long delay
ment such as draglines and trucks and makes the opera- times (>300ms) between rows do not produce as much _,
tion more efficient. Sandia National Laboratories and ICI rock motion as an intermediate row delay time. The rela-
Explosives USA have collaborated since 1987 to develop tionship between row delay timing and rock motion for
methods for computer modeling of coal mine bench intermediate delay timings is not very well understood,
blasting which includes cast blasting. The Sandia com- having had only superficial examination, either in the
puter program, DMC (Distinct_ _Motion _Code), is a dis- field or numerically: A numerical study of this phenome- _ _

crete element
explicit time integration
code thatto track
uses particle
sphericalmotion
elements
resulting
and non is presented in this paper. _ X
D
02
from a blast. A unique feature of DMC is the coupling of _ _-
the rock motion capability with a gas flow capability. The 2 ROW DELAY TIMING AND PERCENT-CAST _ Z
w
code models the flow of the explosive gases outward
from the blastwell. A porous medium is assumed for Optimal row delay timing depends on many variables D
modeling the rock during the gas flow calculation. Spher- including geometry, explosive properties and rock prop- r_O
ical element loads are calculated using the gas flow char- erties. For the following discussion keep in mind that the
acteristics and the porosity of the gas flow model is burden is the material between the explosive row and
modified as the discrete elements move. Input to this either the free face or the next row of explosives (see Fig-
model includes rock properties, geometrical configura- ure 1). Present theory indicates that the optimum timing tt.
tion and explosive equation-of-state parameters. This should allow time for some space to open up in front of -.,O
enables the user to have a wide range of control over the next burden to be shot. This allows each burden to
blast design parameters including explosive type. DMC move freely since it has space into which it can expand.
However, as each burden is blasted it can also assist in
* This work performed at Sandia National Laboratories pushing the burdens that were blasted in front of it, add-
supported by the U.S. Department of Energy under contract no. ing to the momentum of the entire rock mass. Thus, leng
DE-AC04-_D'P00_89.-and also supported by ICI Explosives delay times prevent the burdens from assisting each other i

USA. qz¢/_
• ,-_, ,_c_
...._ :_ to enhance the motion. An optimum delay time is long
 !

, enough to keep the burdens from bunching up while also not requiring any rehandling. Consistent percent-cast dif-
allowing interaction to enhance the motion (Olofsson, ferences of as little as 4% can have a significant impact
1988). This can be likened to a system with a natural fre- on the economics of many surface mining operations.
quency which exhibits a stronger response when the load-
ing frequency and natural frequency are close to each
other. In this paper the same spherical element model was 3 SITE DESCRIPTION
used in a parameter study where only the delay timing
between rows was varied. The example bench blast modeled here occurred at the
Blastwells , Lee Ranch Mine owned by the Santa Fe Pacific Coal

_'Q _ north of Grants, New Mexico. The coal is mined using an

Company. The mine is located approximately 25 miles

_pacing _ over a mile

open pit benchlong and lies
blasting in an where
technique east/west direction.
the pit A
is slightly
cross-section of the pit is shown in Figure 3. The sedi-
Bench Face mentary geology of the site consists of gently dipping
A- j strata where the major coal seams are approximately 27
Burden, Row 2 -I- J" m deep at the western end of the pit and 11 m deep at the
Burden, Row 1 | eastern end. The pit is advanced from west to east with
Face Height the blasting being done in s_tions that are approximately
360 m long. The bench blast studied occurred at the
western end of the pit where the face is approximately 30
m high. A description of the strata at the western end of
the pit is presented in Figure 3 and Table 1. The values
for Youngs' modulus, Poissons' ratio and density were
not actually measured at the site but come from measure-
Figure 1"Definition of Bench Blasting Terms merits on similar materials at other coal mines. The face
and the blastweUs are angled at approximately 15°, and
Comparison between DMC calculations and field data for the spoil pile opposite the face has a 35° angle of repose.
this site is the subject of another paper (Preece, 1993c). The blasting design results from an optimized trade-off
Figure 2 illustrates the definition of percent-cast, a con- between blast casting and preventing the coal from being
cept commonly used to measure blast-induced rock severely damaged or unrecoverable. Higher explosive
motion. The cast-line is parallel to the angle of repose of loads will result in a better cast of the overburden mate-
the spoil pile and intersects the bottom of the face. Per- rial which is more economical because less material
cent-cast refers to that portion of the overburden that is needs to be moved with mechanical equipment. How-
beyond the cast-line and thus resides in its final position, ever, very high explosive loading can result in significant

Original Bench

MuckP A2
Cast-Line Percent-Cast = A1 + A2 x 100

Spoil Pile

Coal Seam

Figure 2: Illustration of Definition of Percent-Cast Based on

Areas A 1 and A 2.
, i

Blastwells (Details Given in Table 2)

irill 0 l0 Scale(m)

_Brown Sandstone Spoil Pile

!: _!i _! i:!i _
_i:i _ i _ _!::_!iil _:::Gray Shale

Gray Sandstone

ii__ Multiple Coal Layers

Figure 3: Schematic of Bench Blast Example Problem

'Pable 1: Material Properties

i

Young's Modulus Poisson's Ratio Specific Gravity

Material Thickness (m) (GPA)
....

Coal Layers 4.48 50.0 0.20 2.10

(multiple)

Gray Sandstone 3.27 40.0 0.12 2.38

Gray Shale 6.05 10.0 0.14 2.52

Brown Sandstone 8.14 40.0 0.12 2.38

.....

Brown Shale 5.49 10.0 0.14 2.52

movement and damage to the coal seam offsetting the sive parameters described in Section 2 have been
economic advantages of higher percent-cast, included in the DMC simulation. The techniques used by
The blast design parameters include burden, spacing, DMC are weU documented in other publications and will
blastwell diameter, explosive type, blastweU angle and not be repeated here. The capabilities and the references
row delay. The blast design parameters are given in Table describing them are given below.
2. The powder factor given in Table 2 is the mass of
explosive per unit volume of rock. It is determined from
the blastweU geometry and explosive density along with 4.1 DMC Capabilities
the burden, spacing and face height. These values are typ-
ical of this type of mine. The powder factor is higher in DMC is a two-dimensional spherical element discrete
the front row because the spacing is half of that on subse- motion code that was originally developed by Taylor and
quentrows. Preece (1989 a&b and 1992) for modeling the motion
associated with rock blasting. Spherical element bulking
mechanisms have been added to allow spherical elements
4 COMPUTER SIMULATION to behave more like rocks (Preece and Taylor, 1990). The
program has been used for modeling confined volume
A DMC model was created to simulate the bench blast blasting associated with oil shale retort blasting (Preece,
just described. All of the geometry, material and explo- 1990 a&b). DMC has been coupled with a gas flow corn-
 Table 2: Blast Design Parameters

Powder
Row No. Burden Spacing Explosive Factor Hole Dia. Hole Angle Row Delay
(m) (m) (Kg/m 3) (ram) (degrees) (ms)
i i i ii

1 4.88 4.57 ANFO 1.3 270.0 15 40 to 500

........

2 thru 7 " 9.14 " 0.63 .... 40 to 500

.........

putation capability so that the explosive loading is auto- three. It also has a significant pile of material next to the
matically treated when an explosive is specified along spoil pile. The dip in the curve in Figure 8, with the bot-
with its physical parameters and equation-of-state tom at 120 ms, can be attributed to the fact that a hump of
(Preece and Knudsen, 1992a),(Preece, 1993a). This past material (120 ms delay profile, Figure 9) does not make it
year DMC has been customized to run in a bench blasting past the cast-line (see Figure 2). Longer delays seem to
environment (Preece and Knudsen, 1992b), and the overcome thisproblem.
importance of spherical element packing angle (Preece, The difference in calculated percent-cast between the
1993b) and row delay timing (Preece,1993c) have been high and low values for this model is only approximately
examined. 7.5% but is large enough to have a significant economic
impact on a mine operation. Careful consideration of the
muck-pile profiles in Figure 9, especially in the area of
4.2 Discrete Element Model the power trench and the associated hump, leads one to
the conclusion that differences in row delay timing pro-
The spherical element model used in this simulation is duce stronger or weaker responses from the system. The
shown in Figure 4. The geologic layers are shown in dif- oscillatory relationship between percent-cast and row
ferent shades of gray. The blastwells are defined geomet- delay timing is probably due to the fact that this is a com-
rically but are not displayed. This spherical element plex dynamic system with many parameters having an
model has a packing angle of 10° which allows signifi- effect. The mine operators have varied row delay timing
cant sphere motion along horizontal bedding planes with- somewhat from blast to blast in search of the optimum,
out much dilatation (Preece, 1993b). The model has 1521 which has been difficult to find. The search may be corn-
spheres and the entire calculation executes in 1191 CPU plicated by the fact that the bench height varies along the
seconds on a SUN SPARCstation 10-41 computer work- pit and optimum timing is probably effected by bench
station, height. Finding an optimum timing by searching along
the curve in Figure 8 one blast at a time also represents a
significant challenge. This type of computer modeling
4.3 Computational Results can complete the 16 simulations required to produce this
curve in less than a day and can thus provide considerable
Figures 5,6 and 7 show the model, with row delay times assistance in blast design.
of I00 ms, at 100 ms, 500 ms and 7 s respectively. The 7
s configuration represents the final muck pile shape. The
graph in Figure 8 illustrates the relationship between row 5 CONCLUSIONS
delay timing and percent-cast derived from 16 different
simulations using this model. Each simulation was per- The ability of DMC to predict differences in muck-pile
formed with a different value of row delay timing and shape and percent-cast as a function of row delay timing
then the final muck pile shape was processed for the cor- has been demonstrated. It was also discovered, during
responding percent-cast. The expectation at the onset of this study, that the relationship between timing and rock
these calculations was a smooth concave downward motion can be quite complex. It was demonstrated that in
curve representing the relationship between row delay accordance with current blasting theory, very short or
and percent-cast. The shape of the derived curve is a bit very long delay times produce the lowest material move-
surprising compared to the expectation. Figure 9 illus- ment. Optimum timing resides somewhere between short
trates five muck pile shapes corresponding to row delay and long delay times where enough time is allowed to
timings of 40 ms, 60 ms, 120 ms, 150 ms and 500 ms. prevent material bunching but the detonations are close
These timings represent the peaks and valleys of the enough to reinforce each other. However, this study also
curve in Figure 8. The 60 ms blast is obviously the best indicates that the timing versus percent-cast curve may
with the deepest trench (commonly referred to as the have a local minimum where one might expect to see a
power trench) to next to the backwall and with the hump local maximum (optimum timing). Thus, an intermediate
beyond the power trench being the furthest out of the row delay timing should not be assumed to produce a per-
3or I

2O

f J I I J
0 10 20 30 40 50 60
Distance (m)

Figure 4: Close-In View of Spherical Element Model.

45

30

I I I I I I
0 15 30 45 60 75 90
Distance (m)

Figure 5: Spherical Element Model at 100 ms. Calculation With 100 ms Delay.
 I I I I I I
0 15 30 45 60 75 90
Distance (m)

Figure 6: Spherical Element Model at 500 ms. Calculation With 100 ms Delay.

30

°_,,_
_15

I I I I I I
0 15 30 45 60 75 90
Distance (m)

Figure 7: Spherical Element Model at 7.0 s. Calculation With 100 ms Delay.

. cerit-cast close 1;ooptimum. Data, Fourth International Symposium on Rock Frag-
A real need exists for further study of the effects of mentation by Blasting, Technical University, Vienna,
row delay timing, especially in the field using data from Austria
full scale bench blasts. This study will continue in the Taylor, L. M. & D. S Preece, 1989a, DMC - A Rigid
future. Body Motion Code for Determining the Interaction of
Multiple Spherical Particles, Sandia National Laborato-
ries, SAND88-3482.
ACKNOWLEDGMENTS Taylor, L. M. & D. S Preece, 1989b, Simulation of Blast-
ing Induced Rock Motion Using Spherical Element
The author wishes to acknowledge the contributions of Models, First U.S. Conference on Discrete Element
personnel from ICI Explosives USA who contributed to Methods, Colorado School of Mines, Golden, Colo-
this work by assisting in acquisition of field data and ana- rado.
lyzing the results. Particular appreciation goes to Steve Taylor, L. M. & D. S Preece, 1992, Simulation of Blasting
Burchell, Scott Scovira, Rusty Zimmerman, Mike Induced Rock Motion Using Spherical Element Mod-
McGill, Joe Strobe, Bob Selby and Steve Woodard. els, Engineering Computations, Vol. 9, No. 2.
Appreciation also goes to the personnel from Lee Ranch
Mine, Ron Covel and Charles Cometti lbr cooperation
and assistance.

REFERENCES

Olofsson, S. O., 1988, Applied Explosives Technology

for Construction and Mining, APPLEX, ,_a'la,Sweden.
Preece, D. S. & L. M. Taylor, 1990, Spherical Element
Bulking Mechanisms for Modeling Blasting Induced

Rock Motion, Third

Fragmentation International
by Blasting, Symposium
Brisbane, on Rock
QL, Australia. __ __ 8. __ _ _" _ '_ "
Preece, D. S., 1990a, Rock Motion Simulation and Pre- '_ _,o
_ _ _ _,o-_ oo=_ _
diction of Porosity Distribution for a Two-Void-Level o =°.-- .._ _.=o
Retort, 23d Annual Oil Shale Symposium, Colorado -."_".,"_ _.
_ _ ,,_
School of Mines, Golden, Colorado. _,.=_
_'"_ _' _,.__'
_ o0 ..- ,-'_ *-, _ o O

Preece, D. S., 1990b, Rock Motion Simulation of Con- _ _ _, ._ e -_ _ _

fined Volume Blasting, 31st U.S. Symposium on Rock _ _ _ _ _.._ ._ _,
Mechanics, Colorado School of Mines, Golden Colo- -,,_ _." _" _"
_ *_o_ o O'_
e_

Preece, D. S., and S. D. Knudsen, 1992a, Coupled Rock _ .. o _ o _ _ _,

Motion and Gas Flow Modeling in Blasting, Eighth _ _ o _-_ = o .
Annual Symposium on Explosives and Blasting _..__=_,_.__
_, _'-_ _.2 _ o 8
t.., _ ._ .,a
Research, International Society of Explosives Engi- r_ o ,., o .i =8,, _ _ _ o_
neers, Orlando Florida. _ _ •._ _ .-._
_ ,8_° o=
_ ._ o_ _"-*_
Preece, D. S., and S. D. Knudsen, 1992b, Computer Mod- _,._ o=_ _ ,_
eling of Gas Flow and Gas Loading of Rock in a Bench = .." ' "",,_ _ _._.
Blasting Environment, 33d U. S Symposium on Rock ,* = _ ._
Mechanics, Santa Fe, New Mexico. ___
L',-" _, _- ,, o_
Preece, D. S., 1993a, Momentum Transfer From Flowing r. = _, _: o o o
Explosive Gases to Spherical Particles During Com- _ z _ __"8 _
puter Simulation of Blasting Induced Rock Motion, = _ . _ _ t_ _ O
Ninth Annual Symposium on Explosives and Blasting g _ _ _ _ '_ _ _ ,,.,'_ 00
Research, International Society of Explosives Engi- _ o-
•-- '__._ ._
_ =8_... ,_. '_ ._._
neers, San Diego, California. _ _ o .0 ,, o _ _ _
Preece, D. S., 1993b, Variation of Spherical Element
Packing Angle and Its Influence on Computer Simula-
tions of Blasting Induced Rock Motion, 2nd Interna-
tional Conference on Discrete Element Methods,
Massachusetts Institute of Technology.
Preece, D. S., 1993c, Coupled Explosive Gas Flow and
Rock Motion With Comparison to Bench Blast Field
 35

34

33'

32-

_
v 31

O 30

29 -e
Q.
28

27

26"

25" , , ' , i , , , i , l
o so loo 15o _oo 2so 300 3so 4oo _o SOD
RowDelay (ms)

Figure 8: Percent-Cast Versus Row Delay.

55

50

45

40
PreblastProfile
35-

E , RowDelay= 40 ms
v 30 - / / RowDelay- 500ms
t" /
25 - / ./ L..--- RowDelay.. 60 ms

RowDelay= 150ms

"%.._TT. " •• " " Delay= 120 ms

10 .- .... . .._.
I .... _.._. ..... .....,,,...;,,_:r
5

0-

"5' 1 ...... I " f I 1 I I I I I I

0 10 20 30 40 50 60 70 80 90 100 110
Distance (m)

Figure 9: Comparison of Muck Piles for Several Different Row Delay Timings.
m m
I I